hyrax-ingest 0.1.0

data/lib/hyrax/ingest/fetcher.rb

@@ -0,0 +1,55 @@
+require 'hyrax/ingest/fetcher/xml_file'
+require 'hyrax/ingest/fetcher/yaml_file'
+require 'hyrax/ingest/fetcher/csv_file'
+require 'hyrax/ingest/fetcher/literal'
+require 'hyrax/ingest/fetcher/premis_event_type'
+require 'hyrax/ingest/fetcher/rdf_uri'
+require 'hyrax/ingest/fetcher/date_time'
+require 'hyrax/ingest/errors'
+
+module Hyrax
+  module Ingest
+    module Fetcher
+      class << self
+        # @return Set The set of all fetcher classes.
+        def all_classes
+          @all_classes ||= Set.new.tap do |all_classes|
+            all_classes << Hyrax::Ingest::Fetcher::XMLFile
+            all_classes << Hyrax::Ingest::Fetcher::YAMLFile
+            all_classes << Hyrax::Ingest::Fetcher::CSVFile
+            all_classes << Hyrax::Ingest::Fetcher::Literal
+            all_classes << Hyrax::Ingest::Fetcher::PremisEventType
+            all_classes << Hyrax::Ingest::Fetcher::RdfUri
+            all_classes << Hyrax::Ingest::Fetcher::DateTime
+          end
+        end
+
+        # @param [string] options The key is the fetcher class name
+        #   and the value is the hash of options to pass to the constructor of the fetcher
+        #   class.
+        # @return An instance of the fetcher class.
+        def factory(class_name, options={})
+          options ||= {}
+          find_class_by_name(class_name).new(options)
+        end
+
+        # @param [String] class_name The stringified class name, with our
+        #   without namespaces.
+        # @raise [Hyrax::Ingest::Errors::UnknownFetcherClass] When there is no
+        #   corresponding fetcher class for the given value of the `class_name`
+        #   param.
+        # @raise [Hyrax::Ingest::Errors::UnknownfetcherClass] When the value of
+        #   `class_name` param is insufficient in determining a fetcher class.
+        # @return [Class] The appropriate fetcher class.
+        def find_class_by_name(class_name)
+          found_classes = all_classes.select do |class_const|
+            (class_const.to_s == class_name) || (class_const.to_s =~ /::#{class_name}/)
+          end
+          raise Hyrax::Ingest::Errors::UnknownFetcherClass.new(class_name, all_classes) if found_classes.count == 0
+          raise Hyrax::Ingest::Errors::AmbiguousFetcherClass.new(class_name, found_classes) if found_classes.count > 1
+          found_classes.first
+        end
+      end
+    end
+  end
+end

data/lib/hyrax/ingest/fetcher/base.rb

@@ -0,0 +1,78 @@
+require 'hyrax/ingest/errors'
+require 'hyrax/ingest/sip'
+require 'interloper'
+require 'hyrax/ingest/has_logger'
+require 'hyrax/ingest/has_report'
+
+
+
+module Hyrax
+  module Ingest
+    module Fetcher
+      class Base
+        include Interloper
+        include HasReport
+        include HasLogger
+
+        # Callback to log missing values that have been marked as required.
+        after(:fetch) do
+          # Use a flag to only report the missing value once.
+          unless @after_fetch_run_once
+            if required? && fetched_value_is_empty?
+              report_missing_required_value
+            end
+            @after_fetch_run_once = true
+          end
+        end
+
+        def initialize(options={})
+          @required = options.delete(:required)
+        end
+
+        # Boolean reader for @required.
+        def required?; !!@required; end
+
+        # Subclasses should override this method with the logic required to
+        # fetch values from a SIP. The overidden method should set
+        # @fetched_value, and return it.
+        # @see Hyrax::Ingest::Fetcher::XMLFile#fetch
+        # @abstract
+        def fetch
+          @fetched_value
+        end
+
+        protected
+
+          # Determines whether the value that was fetched by #fetch is empty
+          # or not. Be default, the fetched value is empty if it is nil, an
+          # empty string, an empty array, an empty hash, or an array
+          # containing any combination of those. Overwrite this method in
+          # subclasses to change the definition of empty in those contexts.
+          # The return value is used for reporting which values are missing,
+          # but required.
+          # @return [Boolean] True if @fetched_value is considered to be empty; false otherwise.
+          def fetched_value_is_empty?
+            Array(@fetched_value).reduce(true) do |all_empty, val|
+              all_empty &&= ( val.nil? || ( val.respond_to?(:empty?) && val.empty? ) )
+            end
+          end
+
+          # Reports occurrences of missing required values.
+          # Subclasses should override this method to provide further detail
+          # by passing an options hash that will be available for reporting.
+          # @see Hyrax::Ingest::Fetcher::XMLFile#report_missing_required_value
+          # @example
+          #     # Provide additional info to the report.
+          #     def report_missing_require_value
+          #       super(foo: "bar")
+          #     end
+          def report_missing_required_value(params={})
+            short_class_name = self.class.to_s.gsub(/.*\:\:/, '')
+            logger.warn "Missing required value from #{short_class_name} with params = #{params}"
+            report.stat[:missing_required_values][self.class] ||= []
+            report.stat[:missing_required_values][self.class] << params
+          end
+      end
+    end
+  end
+end

data/lib/hyrax/ingest/fetcher/csv_file.rb

@@ -0,0 +1,89 @@
+require 'hyrax/ingest/fetcher/base'
+require 'hyrax/ingest/has_sip'
+require 'hyrax/ingest/has_iteration'
+require 'roo'
+
+module Hyrax
+  module Ingest
+    module Fetcher
+      class CSVFile < Base
+        attr_reader :filename, :column, :row, :header_row_number, :row_offset
+
+        include HasSIP
+        include HasIteration
+
+        def initialize(options={})
+          raise ArgumentError, "Required option :filename is missing" unless options.key?(:filename)
+          raise ArgumentError, "Required option :column is missing" unless options.key?(:column)
+          raise ArgumentError, "Required option :row is missing" unless options.key?(:row)
+
+          @filename = options.fetch(:filename)
+          @column = options.fetch(:column, '').to_s.strip.downcase
+          @row = options.fetch(:row, '').to_s.strip.downcase
+          @header_row_number = options.fetch(:header_row_number, 1)
+          @row_offset = options[:row_offset]
+          super
+        end
+
+        def fetch
+          @fetched_value ||= cell_value
+        end
+
+        private
+
+          def roo
+            @roo ||= Roo::CSV.new(sip.find_file_path(filename))
+          end
+
+          def cell_value
+            roo.cell(row_number, column_number)
+          end
+
+          # @return The column number from which to retrieve the cell.
+          def column_number
+            @column_number ||= column_number_from_header || specific_column_number
+            # TODO: custom error
+            raise ArgumentError, "Value for column: option must be a number or a column header; '#{column}' was given." if @column_number.nil?
+            @column_number
+          end
+
+          def column_number_from_header
+            headers.index(column) + 1 if headers.index(column)
+          end
+
+          def specific_column_number
+            column.to_i if string_is_integer?(column)
+          end
+
+          def row_number
+            @row_number ||= next_row || specific_row_number
+            # TODO: custom error
+            raise ArgumentError, "Value for row: option must be a number or the keyword 'next'; #{@row} was given." if @row_number.nil?
+            @row_number + row_offset
+          end
+
+          def next_row
+            iteration + 1 if @row == 'next'
+          end
+
+          def specific_row_number
+            row.to_i if string_is_integer?(row)
+          end
+
+          def row_offset
+            @row_offset || header_row_number
+          end
+
+          def headers
+            @headers ||= roo.row(header_row_number).map(&:to_s).map(&:downcase)
+          end
+
+          def string_is_integer?(str)
+            # remove leading zeros first
+            str.sub!(/^0+/, '')
+            str.to_i.to_s == str.to_s
+          end
+      end
+    end
+  end
+end

data/lib/hyrax/ingest/fetcher/date_time.rb

@@ -0,0 +1,15 @@
+require 'hyrax/ingest/errors'
+require 'hyrax/ingest/sip'
+require 'date'
+
+module Hyrax
+  module Ingest
+    module Fetcher
+      class DateTime < Base
+        def fetch
+          @fetched_value ||= ::DateTime.now
+        end
+      end
+    end
+  end
+end

data/lib/hyrax/ingest/fetcher/literal.rb

@@ -0,0 +1,24 @@
+require 'hyrax/ingest/errors'
+require 'hyrax/ingest/sip'
+
+module Hyrax
+  module Ingest
+    module Fetcher
+      class Literal < Base
+        attr_reader :literal_value
+
+        inherit_callbacks_for :fetch
+
+        def initialize(options={})
+          options = { value: options } unless options.is_a? Hash
+          @literal_value = options.delete(:value)
+          super
+        end
+
+        def fetch
+          @fetched_value = literal_value
+        end
+      end
+    end
+  end
+end

data/lib/hyrax/ingest/fetcher/premis_event_type.rb

@@ -0,0 +1,28 @@
+# This class adhere's to the Hyrax::Ingest::Fetcher::Base interface to work other classes within the Hyrax::Ingest
+# gem. If effectively maps PREMIS Event Type abbreviations to the URI that is then associated with a
+# Hyrax::Preservation::Event model.
+
+require 'hyrax/ingest/errors'
+require 'hyrax/ingest/sip'
+
+module Hyrax
+  module Ingest
+    module Fetcher
+      class PremisEventType < Base
+        inherit_callbacks_for :fetch
+
+        attr_reader :abbr
+
+        def initialize(options={})
+          options = { abbr: options } unless options.is_a? Hash
+          @abbr = options.delete(:abbr)
+          super
+        end
+
+        def fetch
+          @fetched_value ||= Hyrax::Preservation::PremisEventType.find_by_abbr(@abbr).uri
+        end
+      end
+    end
+  end
+end

data/lib/hyrax/ingest/fetcher/rdf_uri.rb

@@ -0,0 +1,21 @@
+require 'rdf'
+
+module Hyrax
+  module Ingest
+    module Fetcher
+      class RdfUri < Base
+        attr_reader :uri_str
+
+        def initialize(options={})
+          options = { uri_str: options } unless options.is_a? Hash
+          @uri_str = options.delete(:uri_str)
+          super
+        end
+
+        def fetch
+          @fetched_value ||= ::RDF::URI.new(uri_str)
+        end
+      end
+    end
+  end
+end

data/lib/hyrax/ingest/fetcher/xml_file.rb

@@ -0,0 +1,54 @@
+require 'hyrax/ingest/fetcher/base'
+require 'hyrax/ingest/has_sip'
+require 'nokogiri'
+require 'hyrax/ingest/reporting'
+require 'interloper'
+
+module Hyrax
+  module Ingest
+    module Fetcher
+      class XMLFile < Base
+        include HasSIP
+
+        attr_reader :filename, :xpath, :default, :fetched_value
+
+        inherit_callbacks_after :fetch
+
+        def initialize(options={})
+          raise ArgumentError, "Required option :filename is missing" unless options.key?(:filename)
+          raise ArgumentError, "Required option :xpath is missing" unless options.key?(:xpath)
+          @filename = options[:filename]
+          @xpath = options[:xpath]
+          @default = options[:default] || []
+          super
+        end
+
+        # Overrides Hyrax::Ingest::Fetcher::Base#fetch
+        # @return [String] The value fetched from the XML file
+        def fetch
+          @fetched_value ||= begin
+            fetched_from_xml = noko.xpath(xpath).map(&:text)
+            fetched_from_xml.empty? ? default : fetched_from_xml
+          end
+        end
+
+        protected
+
+          # Overrides
+          # Hyrax::Ingest::Fetcher::Base#report_missing_required_value,
+          # passing filename and xpath inforamtion. The report summary has
+          # specific logic to look for these.
+          def report_missing_required_value
+            super(filename: filename, xpath: xpath)
+          end
+
+          def noko
+            @noko ||= Nokogiri::XML(sip.read_file(filename)).tap do |n|
+              # TODO: allow using namespaces instead of blindly removing them.
+              n.remove_namespaces!
+            end
+          end
+      end
+    end
+  end
+end

data/lib/hyrax/ingest/fetcher/yaml_file.rb

@@ -0,0 +1,36 @@
+require 'hyrax/ingest/fetcher/base'
+require 'hyrax/ingest/has_sip'
+require 'yaml'
+
+module Hyrax
+  module Ingest
+    module Fetcher
+      class YAMLFile < Base
+        include HasSIP
+
+        inherit_callbacks_for :fetch
+
+        attr_reader :filename, :yaml_path
+
+        def initialize(options={})
+          raise ArgumentError, "Required option :filename is missing" unless options.key?(:filename)
+          raise ArgumentError, "Required option :yaml_path is missing" unless options.key?(:yaml_path)
+          @filename = options[:filename]
+          @yaml_path = options[:yaml_path]
+          super
+        end
+
+        def fetch
+          @fetched_value ||= yaml.dig(*yaml_path)
+        end
+
+        private
+
+          # @return [Hash, Array, String] The YAML parsed into Ruby objects.
+          def yaml
+            @yaml ||= YAML.load sip.read_file(filename)
+          end
+      end
+    end
+  end
+end

data/lib/hyrax/ingest/has_depositor.rb

@@ -0,0 +1,13 @@
+# Simple interface for injecting a Hyrax::Ingest::SIP dependency.
+require 'hyrax/ingest/sip'
+require 'hyrax/ingest/errors'
+
+module Hyrax
+  module Ingest
+    module HasDepositor
+      def HasDepositor.included(mod)
+        attr_accessor :depositor
+      end
+    end
+  end
+end

data/lib/hyrax/ingest/has_iteration.rb

@@ -0,0 +1,18 @@
+# Simple interface for injecting a Hyrax::Ingest::SIP dependency.
+
+require 'hyrax/ingest/sip'
+require 'hyrax/ingest/errors'
+
+module Hyrax
+  module Ingest
+    module HasIteration
+      def iteration=(iteration)
+        @iteration = iteration.to_i
+      end
+
+      def iteration
+        @iteration ||= 0
+      end
+    end
+  end
+end

data/lib/hyrax/ingest/has_logger.rb

@@ -0,0 +1,29 @@
+require 'logger'
+
+module Hyrax
+  module Ingest
+    module HasLogger
+      def logger
+        @logger ||= Logger.new(HasLogger.config.default_log_to)
+      end
+
+      def logger=(logger)
+        raise Hyrax::Ingest::Errors::InvalidLogger unless logger.is_a? Logger
+        @logger = logger
+      end
+
+      class << self
+        def config
+          @config ||= Configuration.new
+        end
+      end
+
+      class Configuration
+        attr_accessor :default_log_to
+        def initialize
+          @default_log_to = STDOUT
+        end
+      end
+    end
+  end
+end